Allen Wittenauer wrote:
On Dec 14, 2015, at 9:43 PM, Josh Elser<[email protected]> wrote:
Augh! I forgot we still did this :). It's likely a holdover from old-days.
Completely agree the generated docs should be under target/
I've been trying to rerun it myself, happy to see you beat me to the punch. I
appreciate you looking into it, Allen.
No problem. It’s always interesting to see where Yetus falls down.
haha. (…and, FWIW, as I play around with Accumulo, I can see a few more gaps…
*sigh*)
BTW, I *do* have a partial workaround for you and I can’t believe I
forgot to try it last night: Turn off javadoc! Using --plugins=“all,-javadoc”
of course skips the javadoc tests, but the rest of test-patch seems to be
working as expected. This way you can get at least some of the bits working
while the work of moving the doc generation to a ‘safe’ location proceeds.
ooo, I didn't realize I could negate plugins. I had just specified a
positive list that didn't include javadoc :D. I've been slowly kicking
the tires. Really nice stuff.
One other thing I noticed is that findbugs, checkstyle, mvn site, etc,
etc, run on *every* invocation. That’s either good or bad, depending upon how
much control one wants over the build. For Yetus, it definitely has a
significant run time cost.
A follow-on question I was going to ask you :).
It seems to me that the mvninstall "invocation" (sorry, still learning terminology)
should also include "-Dfindbugs.skip=true -Dcheckstyle.skip=true". My understanding of
Maven convention is that is it expected that these sorts of checks would run as a part of `mvn
verify` (I forget the exact lifecycle phase they bind to, just that it's included invoking
something at-or-after verify). Would it make sense to also add these arguments for the pre-patch
and patch install?
Yes. There’s a few more skips that need to get added too based upon
what I’m seeing in Accumulo. Would you mind filing a JIRA? Thanks!
Happy to! Thanks again for your help.
